From 2824bc9328467127083c1325f54b67d298c333b2 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Thu, 20 Oct 2005 10:56:41 +0200 Subject: [PATCH] fix try_module_get race in elevator_find This patch removes try_module_get race in elevator_find. try_module_get should always be called with the spinlock protecting what the module init/cleanup routines register/unregister to held. In the case of elevators, we should be holding elv_list to avoid it going away between spin_unlock_irq and try_module_get. Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- drivers/block/elevator.c | 24 ++++++++++++++---------- 1 file changed, 14 insertions(+), 10 deletions(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index 98f0126a2deb..4144f30d82a9 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -97,7 +97,6 @@ static struct elevator_type *elevator_find(const char *name) struct elevator_type *e = NULL; struct list_head *entry; - spin_lock_irq(&elv_list_lock); list_for_each(entry, &elv_list) { struct elevator_type *__e; @@ -108,7 +107,6 @@ static struct elevator_type *elevator_find(const char *name) break; } } - spin_unlock_irq(&elv_list_lock); return e; } @@ -120,12 +118,15 @@ static void elevator_put(struct elevator_type *e) static struct elevator_type *elevator_get(const char *name) { - struct elevator_type *e = elevator_find(name); + struct elevator_type *e; - if (!e) - return NULL; - if (!try_module_get(e->elevator_owner)) - return NULL; + spin_lock_irq(&elv_list_lock); + + e = elevator_find(name); + if (e && !try_module_get(e->elevator_owner)) + e = NULL; + + spin_unlock_irq(&elv_list_lock); return e; } @@ -153,11 +154,15 @@ static char chosen_elevator[16]; static void elevator_setup_default(void) { + struct elevator_type *e; + /* * check if default is set and exists */ - if (chosen_elevator[0] && elevator_find(chosen_elevator)) + if (chosen_elevator[0] && (e = elevator_get(chosen_elevator))) { + elevator_put(e); return; + } #if defined(CONFIG_IOSCHED_AS) strcpy(chosen_elevator, "anticipatory"); @@ -555,10 +560,9 @@ void elv_unregister_queue(struct request_queue *q) int elv_register(struct elevator_type *e) { + spin_lock_irq(&elv_list_lock); if (elevator_find(e->elevator_name)) BUG(); - - spin_lock_irq(&elv_list_lock); list_add_tail(&e->list, &elv_list); spin_unlock_irq(&elv_list_lock); -- cgit v1.2.3 From 8922e16cf6269e668123acb1ae1fdc62b7a3a4fc Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Thu, 20 Oct 2005 16:23:44 +0200 Subject: [PATCH] 01/05 Implement generic dispatch queue Implements generic dispatch queue which can replace all dispatch queues implemented by each iosched. This reduces code duplication, eases enforcing semantics over dispatch queue, and simplifies specific ioscheds. Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- drivers/block/elevator.c | 242 ++++++++++++++++++++++++++++++++-------------- drivers/block/ll_rw_blk.c | 23 +++-- include/linux/blkdev.h | 17 ++-- include/linux/elevator.h | 16 ++- 4 files changed, 201 insertions(+), 97 deletions(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index 4144f30d82a9..a27555908d35 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -40,6 +40,11 @@ static DEFINE_SPINLOCK(elv_list_lock); static LIST_HEAD(elv_list); +static inline sector_t rq_last_sector(struct request *rq) +{ + return rq->sector + rq->nr_sectors; +} + /* * can we safely merge with this request? */ @@ -143,6 +148,9 @@ static int elevator_attach(request_queue_t *q, struct elevator_type *e, INIT_LIST_HEAD(&q->queue_head); q->last_merge = NULL; q->elevator = eq; + q->last_sector = 0; + q->boundary_rq = NULL; + q->max_back_kb = 0; if (eq->ops->elevator_init_fn) ret = eq->ops->elevator_init_fn(q, eq); @@ -225,6 +233,48 @@ void elevator_exit(elevator_t *e) kfree(e); } +/* + * Insert rq into dispatch queue of q. Queue lock must be held on + * entry. If sort != 0, rq is sort-inserted; otherwise, rq will be + * appended to the dispatch queue. To be used by specific elevators. + */ +void elv_dispatch_insert(request_queue_t *q, struct request *rq, int sort) +{ + sector_t boundary; + unsigned max_back; + struct list_head *entry; + + if (!sort) { + /* Specific elevator is performing sort. Step away. */ + q->last_sector = rq_last_sector(rq); + q->boundary_rq = rq; + list_add_tail(&rq->queuelist, &q->queue_head); + return; + } + + boundary = q->last_sector; + max_back = q->max_back_kb * 2; + boundary = boundary > max_back ? boundary - max_back : 0; + + list_for_each_prev(entry, &q->queue_head) { + struct request *pos = list_entry_rq(entry); + + if (pos->flags & (REQ_SOFTBARRIER|REQ_HARDBARRIER|REQ_STARTED)) + break; + if (rq->sector >= boundary) { + if (pos->sector < boundary) + continue; + } else { + if (pos->sector >= boundary) + break; + } + if (rq->sector >= pos->sector) + break; + } + + list_add(&rq->queuelist, entry); +} + int elv_merge(request_queue_t *q, struct request **req, struct bio *bio) { elevator_t *e = q->elevator; @@ -255,13 +305,7 @@ void elv_merge_requests(request_queue_t *q, struct request *rq, e->ops->elevator_merge_req_fn(q, rq, next); } -/* - * For careful internal use by the block layer. Essentially the same as - * a requeue in that it tells the io scheduler that this request is not - * active in the driver or hardware anymore, but we don't want the request - * added back to the scheduler. Function is not exported. - */ -void elv_deactivate_request(request_queue_t *q, struct request *rq) +void elv_requeue_request(request_queue_t *q, struct request *rq) { elevator_t *e = q->elevator; @@ -269,19 +313,14 @@ void elv_deactivate_request(request_queue_t *q, struct request *rq) * it already went through dequeue, we need to decrement the * in_flight count again */ - if (blk_account_rq(rq)) + if (blk_account_rq(rq)) { q->in_flight--; + if (blk_sorted_rq(rq) && e->ops->elevator_deactivate_req_fn) + e->ops->elevator_deactivate_req_fn(q, rq); + } rq->flags &= ~REQ_STARTED; - if (e->ops->elevator_deactivate_req_fn) - e->ops->elevator_deactivate_req_fn(q, rq); -} - -void elv_requeue_request(request_queue_t *q, struct request *rq) -{ - elv_deactivate_request(q, rq); - /* * if this is the flush, requeue the original instead and drop the flush */ @@ -290,55 +329,89 @@ void elv_requeue_request(request_queue_t *q, struct request *rq) rq = rq->end_io_data; } - /* - * the request is prepped and may have some resources allocated. - * allowing unprepped requests to pass this one may cause resource - * deadlock. turn on softbarrier. - */ - rq->flags |= REQ_SOFTBARRIER; - - /* - * if iosched has an explicit requeue hook, then use that. otherwise - * just put the request at the front of the queue - */ - if (q->elevator->ops->elevator_requeue_req_fn) - q->elevator->ops->elevator_requeue_req_fn(q, rq); - else - __elv_add_request(q, rq, ELEVATOR_INSERT_FRONT, 0); + __elv_add_request(q, rq, ELEVATOR_INSERT_FRONT, 0); } void __elv_add_request(request_queue_t *q, struct request *rq, int where, int plug) { - /* - * barriers implicitly indicate back insertion - */ - if (rq->flags & (REQ_SOFTBARRIER | REQ_HARDBARRIER) && - where == ELEVATOR_INSERT_SORT) - where = ELEVATOR_INSERT_BACK; + if (rq->flags & (REQ_SOFTBARRIER | REQ_HARDBARRIER)) { + /* + * barriers implicitly indicate back insertion + */ + if (where == ELEVATOR_INSERT_SORT) + where = ELEVATOR_INSERT_BACK; + + /* + * this request is scheduling boundary, update last_sector + */ + if (blk_fs_request(rq)) { + q->last_sector = rq_last_sector(rq); + q->boundary_rq = rq; + } + } if (plug) blk_plug_device(q); rq->q = q; - if (!test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags)) { - q->elevator->ops->elevator_add_req_fn(q, rq, where); - - if (blk_queue_plugged(q)) { - int nrq = q->rq.count[READ] + q->rq.count[WRITE] - - q->in_flight; - - if (nrq >= q->unplug_thresh) - __generic_unplug_device(q); - } - } else + if (unlikely(test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags))) { /* * if drain is set, store the request "locally". when the drain * is finished, the requests will be handed ordered to the io * scheduler */ list_add_tail(&rq->queuelist, &q->drain_list); + return; + } + + switch (where) { + case ELEVATOR_INSERT_FRONT: + rq->flags |= REQ_SOFTBARRIER; + + list_add(&rq->queuelist, &q->queue_head); + break; + + case ELEVATOR_INSERT_BACK: + rq->flags |= REQ_SOFTBARRIER; + + while (q->elevator->ops->elevator_dispatch_fn(q, 1)) + ; + list_add_tail(&rq->queuelist, &q->queue_head); + /* + * We kick the queue here for the following reasons. + * - The elevator might have returned NULL previously + * to delay requests and returned them now. As the + * queue wasn't empty before this request, ll_rw_blk + * won't run the queue on return, resulting in hang. + * - Usually, back inserted requests won't be merged + * with anything. There's no point in delaying queue + * processing. + */ + blk_remove_plug(q); + q->request_fn(q); + break; + + case ELEVATOR_INSERT_SORT: + BUG_ON(!blk_fs_request(rq)); + rq->flags |= REQ_SORTED; + q->elevator->ops->elevator_add_req_fn(q, rq); + break; + + default: + printk(KERN_ERR "%s: bad insertion point %d\n", + __FUNCTION__, where); + BUG(); + } + + if (blk_queue_plugged(q)) { + int nrq = q->rq.count[READ] + q->rq.count[WRITE] + - q->in_flight; + + if (nrq >= q->unplug_thresh) + __generic_unplug_device(q); + } } void elv_add_request(request_queue_t *q, struct request *rq, int where, @@ -353,13 +426,19 @@ void elv_add_request(request_queue_t *q, struct request *rq, int where, static inline struct request *__elv_next_request(request_queue_t *q) { - struct request *rq = q->elevator->ops->elevator_next_req_fn(q); + struct request *rq; + + if (unlikely(list_empty(&q->queue_head) && + !q->elevator->ops->elevator_dispatch_fn(q, 0))) + return NULL; + + rq = list_entry_rq(q->queue_head.next); /* * if this is a barrier write and the device has to issue a * flush sequence to support it, check how far we are */ - if (rq && blk_fs_request(rq) && blk_barrier_rq(rq)) { + if (blk_fs_request(rq) && blk_barrier_rq(rq)) { BUG_ON(q->ordered == QUEUE_ORDERED_NONE); if (q->ordered == QUEUE_ORDERED_FLUSH && @@ -376,16 +455,34 @@ struct request *elv_next_request(request_queue_t *q) int ret; while ((rq = __elv_next_request(q)) != NULL) { - /* - * just mark as started even if we don't start it, a request - * that has been delayed should not be passed by new incoming - * requests - */ - rq->flags |= REQ_STARTED; + if (!(rq->flags & REQ_STARTED)) { + elevator_t *e = q->elevator; + + /* + * This is the first time the device driver + * sees this request (possibly after + * requeueing). Notify IO scheduler. + */ + if (blk_sorted_rq(rq) && + e->ops->elevator_activate_req_fn) + e->ops->elevator_activate_req_fn(q, rq); + + /* + * just mark as started even if we don't start + * it, a request that has been delayed should + * not be passed by new incoming requests + */ + rq->flags |= REQ_STARTED; + } if (rq == q->last_merge) q->last_merge = NULL; + if (!q->boundary_rq || q->boundary_rq == rq) { + q->last_sector = rq_last_sector(rq); + q->boundary_rq = NULL; + } + if ((rq->flags & REQ_DONTPREP) || !q->prep_rq_fn) break; @@ -396,9 +493,9 @@ struct request *elv_next_request(request_queue_t *q) /* * the request may have been (partially) prepped. * we need to keep this request in the front to - * avoid resource deadlock. turn on softbarrier. + * avoid resource deadlock. REQ_STARTED will + * prevent other fs requests from passing this one. */ - rq->flags |= REQ_SOFTBARRIER; rq = NULL; break; } else if (ret == BLKPREP_KILL) { @@ -421,16 +518,16 @@ struct request *elv_next_request(request_queue_t *q) return rq; } -void elv_remove_request(request_queue_t *q, struct request *rq) +void elv_dequeue_request(request_queue_t *q, struct request *rq) { - elevator_t *e = q->elevator; + BUG_ON(list_empty(&rq->queuelist)); + + list_del_init(&rq->queuelist); /* * the time frame between a request being removed from the lists * and to it is freed is accounted as io that is in progress at - * the driver side. note that we only account requests that the - * driver has seen (REQ_STARTED set), to avoid false accounting - * for request-request merges + * the driver side. */ if (blk_account_rq(rq)) q->in_flight++; @@ -444,19 +541,19 @@ void elv_remove_request(request_queue_t *q, struct request *rq) */ if (rq == q->last_merge) q->last_merge = NULL; - - if (e->ops->elevator_remove_req_fn) - e->ops->elevator_remove_req_fn(q, rq); } int elv_queue_empty(request_queue_t *q) { elevator_t *e = q->elevator; + if (!list_empty(&q->queue_head)) + return 0; + if (e->ops->elevator_queue_empty_fn) return e->ops->elevator_queue_empty_fn(q); - return list_empty(&q->queue_head); + return 1; } struct request *elv_latter_request(request_queue_t *q, struct request *rq) @@ -528,11 +625,11 @@ void elv_completed_request(request_queue_t *q, struct request *rq) /* * request is released from the driver, io must be done */ - if (blk_account_rq(rq)) + if (blk_account_rq(rq)) { q->in_flight--; - - if (e->ops->elevator_completed_req_fn) - e->ops->elevator_completed_req_fn(q, rq); + if (blk_sorted_rq(rq) && e->ops->elevator_completed_req_fn) + e->ops->elevator_completed_req_fn(q, rq); + } } int elv_register_queue(struct request_queue *q) @@ -705,11 +802,12 @@ ssize_t elv_iosched_show(request_queue_t *q, char *name) return len; } +EXPORT_SYMBOL(elv_dispatch_insert); EXPORT_SYMBOL(elv_add_request); EXPORT_SYMBOL(__elv_add_request); EXPORT_SYMBOL(elv_requeue_request); EXPORT_SYMBOL(elv_next_request); -EXPORT_SYMBOL(elv_remove_request); +EXPORT_SYMBOL(elv_dequeue_request); EXPORT_SYMBOL(elv_queue_empty); EXPORT_SYMBOL(elv_completed_request); EXPORT_SYMBOL(elevator_exit); diff --git a/drivers/block/ll_rw_blk.c b/drivers/block/ll_rw_blk.c index 4e2b1b06b411..d2a66fd309c3 100644 --- a/drivers/block/ll_rw_blk.c +++ b/drivers/block/ll_rw_blk.c @@ -353,6 +353,8 @@ static void blk_pre_flush_end_io(struct request *flush_rq) struct request *rq = flush_rq->end_io_data; request_queue_t *q = rq->q; + elv_completed_request(q, flush_rq); + rq->flags |= REQ_BAR_PREFLUSH; if (!flush_rq->errors) @@ -369,6 +371,8 @@ static void blk_post_flush_end_io(struct request *flush_rq) struct request *rq = flush_rq->end_io_data; request_queue_t *q = rq->q; + elv_completed_request(q, flush_rq); + rq->flags |= REQ_BAR_POSTFLUSH; q->end_flush_fn(q, flush_rq); @@ -408,8 +412,6 @@ struct request *blk_start_pre_flush(request_queue_t *q, struct request *rq) if (!list_empty(&rq->queuelist)) blkdev_dequeue_request(rq); - elv_deactivate_request(q, rq); - flush_rq->end_io_data = rq; flush_rq->end_io = blk_pre_flush_end_io; @@ -1040,6 +1042,7 @@ EXPORT_SYMBOL(blk_queue_invalidate_tags); static char *rq_flags[] = { "REQ_RW", "REQ_FAILFAST", + "REQ_SORTED", "REQ_SOFTBARRIER", "REQ_HARDBARRIER", "REQ_CMD", @@ -2456,6 +2459,8 @@ static void __blk_put_request(request_queue_t *q, struct request *req) if (unlikely(--req->ref_count)) return; + elv_completed_request(q, req); + req->rq_status = RQ_INACTIVE; req->rl = NULL; @@ -2466,8 +2471,6 @@ static void __blk_put_request(request_queue_t *q, struct request *req) if (rl) { int rw = rq_data_dir(req); - elv_completed_request(q, req); - BUG_ON(!list_empty(&req->queuelist)); blk_free_request(q, req); @@ -2477,14 +2480,14 @@ static void __blk_put_request(request_queue_t *q, struct request *req) void blk_put_request(struct request *req) { + unsigned long flags; + request_queue_t *q = req->q; + /* - * if req->rl isn't set, this request didnt originate from the - * block layer, so it's safe to just disregard it + * Gee, IDE calls in w/ NULL q. Fix IDE and remove the + * following if (q) test. */ - if (req->rl) { - unsigned long flags; - request_queue_t *q = req->q; - + if (q) { spin_lock_irqsave(q->queue_lock, flags); __blk_put_request(q, req); spin_unlock_irqrestore(q->queue_lock, flags); diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index efdc9b5bc05c..2c7b9154927a 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -203,6 +203,7 @@ struct request { enum rq_flag_bits { __REQ_RW, /* not set, read. set, write */ __REQ_FAILFAST, /* no low level driver retries */ + __REQ_SORTED, /* elevator knows about this request */ __REQ_SOFTBARRIER, /* may not be passed by ioscheduler */ __REQ_HARDBARRIER, /* may not be passed by drive either */ __REQ_CMD, /* is a regular fs rw request */ @@ -235,6 +236,7 @@ enum rq_flag_bits { #define REQ_RW (1 << __REQ_RW) #define REQ_FAILFAST (1 << __REQ_FAILFAST) +#define REQ_SORTED (1 << __REQ_SORTED) #define REQ_SOFTBARRIER (1 << __REQ_SOFTBARRIER) #define REQ_HARDBARRIER (1 << __REQ_HARDBARRIER) #define REQ_CMD (1 << __REQ_CMD) @@ -332,6 +334,13 @@ struct request_queue prepare_flush_fn *prepare_flush_fn; end_flush_fn *end_flush_fn; + /* + * Dispatch queue sorting + */ + sector_t last_sector; + struct request *boundary_rq; + unsigned int max_back_kb; + /* * Auto-unplugging state */ @@ -454,6 +463,7 @@ enum { #define blk_pm_request(rq) \ ((rq)->flags & (REQ_PM_SUSPEND | REQ_PM_RESUME)) +#define blk_sorted_rq(rq) ((rq)->flags & REQ_SORTED) #define blk_barrier_rq(rq) ((rq)->flags & REQ_HARDBARRIER) #define blk_barrier_preflush(rq) ((rq)->flags & REQ_BAR_PREFLUSH) #define blk_barrier_postflush(rq) ((rq)->flags & REQ_BAR_POSTFLUSH) @@ -611,12 +621,7 @@ extern void end_request(struct request *req, int uptodate); static inline void blkdev_dequeue_request(struct request *req) { - BUG_ON(list_empty(&req->queuelist)); - - list_del_init(&req->queuelist); - - if (req->rl) - elv_remove_request(req->q, req); + elv_dequeue_request(req->q, req); } /* diff --git a/include/linux/elevator.h b/include/linux/elevator.h index ea6bbc2d7407..76f4f6920744 100644 --- a/include/linux/elevator.h +++ b/include/linux/elevator.h @@ -8,18 +8,17 @@ typedef void (elevator_merge_req_fn) (request_queue_t *, struct request *, struc typedef void (elevator_merged_fn) (request_queue_t *, struct request *); -typedef struct request *(elevator_next_req_fn) (request_queue_t *); +typedef int (elevator_dispatch_fn) (request_queue_t *, int); -typedef void (elevator_add_req_fn) (request_queue_t *, struct request *, int); +typedef void (elevator_add_req_fn) (request_queue_t *, struct request *); typedef int (elevator_queue_empty_fn) (request_queue_t *); -typedef void (elevator_remove_req_fn) (request_queue_t *, struct request *); -typedef void (elevator_requeue_req_fn) (request_queue_t *, struct request *); typedef struct request *(elevator_request_list_fn) (request_queue_t *, struct request *); typedef void (elevator_completed_req_fn) (request_queue_t *, struct request *); typedef int (elevator_may_queue_fn) (request_queue_t *, int, struct bio *); typedef int (elevator_set_req_fn) (request_queue_t *, struct request *, struct bio *, int); typedef void (elevator_put_req_fn) (request_queue_t *, struct request *); +typedef void (elevator_activate_req_fn) (request_queue_t *, struct request *); typedef void (elevator_deactivate_req_fn) (request_queue_t *, struct request *); typedef int (elevator_init_fn) (request_queue_t *, elevator_t *); @@ -31,10 +30,9 @@ struct elevator_ops elevator_merged_fn *elevator_merged_fn; elevator_merge_req_fn *elevator_merge_req_fn; - elevator_next_req_fn *elevator_next_req_fn; + elevator_dispatch_fn *elevator_dispatch_fn; elevator_add_req_fn *elevator_add_req_fn; - elevator_remove_req_fn *elevator_remove_req_fn; - elevator_requeue_req_fn *elevator_requeue_req_fn; + elevator_activate_req_fn *elevator_activate_req_fn; elevator_deactivate_req_fn *elevator_deactivate_req_fn; elevator_queue_empty_fn *elevator_queue_empty_fn; @@ -81,15 +79,15 @@ struct elevator_queue /* * block elevator interface */ +extern void elv_dispatch_insert(request_queue_t *, struct request *, int); extern void elv_add_request(request_queue_t *, struct request *, int, int); extern void __elv_add_request(request_queue_t *, struct request *, int, int); extern int elv_merge(request_queue_t *, struct request **, struct bio *); extern void elv_merge_requests(request_queue_t *, struct request *, struct request *); extern void elv_merged_request(request_queue_t *, struct request *); -extern void elv_remove_request(request_queue_t *, struct request *); +extern void elv_dequeue_request(request_queue_t *, struct request *); extern void elv_requeue_request(request_queue_t *, struct request *); -extern void elv_deactivate_request(request_queue_t *, struct request *); extern int elv_queue_empty(request_queue_t *); extern struct request *elv_next_request(struct request_queue *q); extern struct request *elv_former_request(request_queue_t *, struct request *); -- cgit v1.2.3 From 1b47f531e244e339b648dfff071c086f936e49e1 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Thu, 20 Oct 2005 16:37:00 +0200 Subject: [PATCH] generic dispatch fixes - Split elv_dispatch_insert() into two functions - Rename rq_last_sector() to rq_end_sector() Signed-off-by: Jens Axboe --- drivers/block/elevator.c | 33 ++++++++++++--------------------- include/linux/blkdev.h | 13 ++++++++++++- include/linux/elevator.h | 4 +++- 3 files changed, 27 insertions(+), 23 deletions(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index a27555908d35..237e43d760c8 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -40,11 +40,6 @@ static DEFINE_SPINLOCK(elv_list_lock); static LIST_HEAD(elv_list); -static inline sector_t rq_last_sector(struct request *rq) -{ - return rq->sector + rq->nr_sectors; -} - /* * can we safely merge with this request? */ @@ -148,7 +143,7 @@ static int elevator_attach(request_queue_t *q, struct elevator_type *e, INIT_LIST_HEAD(&q->queue_head); q->last_merge = NULL; q->elevator = eq; - q->last_sector = 0; + q->end_sector = 0; q->boundary_rq = NULL; q->max_back_kb = 0; @@ -233,29 +228,25 @@ void elevator_exit(elevator_t *e) kfree(e); } +void elv_dispatch_insert_tail(request_queue_t *q, struct request *rq) +{ +} + /* * Insert rq into dispatch queue of q. Queue lock must be held on * entry. If sort != 0, rq is sort-inserted; otherwise, rq will be * appended to the dispatch queue. To be used by specific elevators. */ -void elv_dispatch_insert(request_queue_t *q, struct request *rq, int sort) +void elv_dispatch_sort(request_queue_t *q, struct request *rq) { sector_t boundary; unsigned max_back; struct list_head *entry; - if (!sort) { - /* Specific elevator is performing sort. Step away. */ - q->last_sector = rq_last_sector(rq); - q->boundary_rq = rq; - list_add_tail(&rq->queuelist, &q->queue_head); - return; - } - - boundary = q->last_sector; + boundary = q->end_sector; max_back = q->max_back_kb * 2; boundary = boundary > max_back ? boundary - max_back : 0; - + list_for_each_prev(entry, &q->queue_head) { struct request *pos = list_entry_rq(entry); @@ -343,10 +334,10 @@ void __elv_add_request(request_queue_t *q, struct request *rq, int where, where = ELEVATOR_INSERT_BACK; /* - * this request is scheduling boundary, update last_sector + * this request is scheduling boundary, update end_sector */ if (blk_fs_request(rq)) { - q->last_sector = rq_last_sector(rq); + q->end_sector = rq_end_sector(rq); q->boundary_rq = rq; } } @@ -479,7 +470,7 @@ struct request *elv_next_request(request_queue_t *q) q->last_merge = NULL; if (!q->boundary_rq || q->boundary_rq == rq) { - q->last_sector = rq_last_sector(rq); + q->end_sector = rq_end_sector(rq); q->boundary_rq = NULL; } @@ -802,7 +793,7 @@ ssize_t elv_iosched_show(request_queue_t *q, char *name) return len; } -EXPORT_SYMBOL(elv_dispatch_insert); +EXPORT_SYMBOL(elv_dispatch_sort); EXPORT_SYMBOL(elv_add_request); EXPORT_SYMBOL(__elv_add_request); EXPORT_SYMBOL(elv_requeue_request); diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 2c7b9154927a..a2e496f285a4 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -337,7 +337,7 @@ struct request_queue /* * Dispatch queue sorting */ - sector_t last_sector; + sector_t end_sector; struct request *boundary_rq; unsigned int max_back_kb; @@ -624,6 +624,17 @@ static inline void blkdev_dequeue_request(struct request *req) elv_dequeue_request(req->q, req); } +/* + * This should be in elevator.h, but that requires pulling in rq and q + */ +static inline void elv_dispatch_add_tail(struct request_queue *q, + struct request *rq) +{ + q->end_sector = rq_end_sector(rq); + q->boundary_rq = rq; + list_add_tail(&rq->queuelist, &q->queue_head); +} + /* * Access functions for manipulating queue properties */ diff --git a/include/linux/elevator.h b/include/linux/elevator.h index 76f4f6920744..f9b5ec38121a 100644 --- a/include/linux/elevator.h +++ b/include/linux/elevator.h @@ -79,7 +79,7 @@ struct elevator_queue /* * block elevator interface */ -extern void elv_dispatch_insert(request_queue_t *, struct request *, int); +extern void elv_dispatch_sort(request_queue_t *, struct request *); extern void elv_add_request(request_queue_t *, struct request *, int, int); extern void __elv_add_request(request_queue_t *, struct request *, int, int); extern int elv_merge(request_queue_t *, struct request **, struct bio *); @@ -140,4 +140,6 @@ enum { ELV_MQUEUE_MUST, }; +#define rq_end_sector(rq) ((rq)->sector + (rq)->nr_sectors) + #endif -- cgit v1.2.3 From d9ebb192aa13a026edc6faff137dcb14f2c91731 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Thu, 20 Oct 2005 16:37:38 +0200 Subject: [PATCH] elevator: leftover function declaration Signed-off-by: Jens Axboe --- drivers/block/elevator.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index 237e43d760c8..29d6c8237ab3 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -228,10 +228,6 @@ void elevator_exit(elevator_t *e) kfree(e); } -void elv_dispatch_insert_tail(request_queue_t *q, struct request *rq) -{ -} - /* * Insert rq into dispatch queue of q. Queue lock must be held on * entry. If sort != 0, rq is sort-inserted; otherwise, rq will be -- cgit v1.2.3 From 06b86245c052963029bfd9020ca1f08ceb66f85a Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Thu, 20 Oct 2005 16:46:23 +0200 Subject: [PATCH] 03/05 move last_merge handlin into generic elevator code Currently, both generic elevator code and specific ioscheds participate in the management and usage of last_merge. This and the following patches move last_merge handling into generic elevator code. Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- drivers/block/elevator.c | 43 ++++++++++++++++++------------------------- include/linux/blkdev.h | 3 +++ 2 files changed, 21 insertions(+), 25 deletions(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index 29d6c8237ab3..415144372c75 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -83,15 +83,6 @@ inline int elv_try_merge(struct request *__rq, struct bio *bio) } EXPORT_SYMBOL(elv_try_merge); -inline int elv_try_last_merge(request_queue_t *q, struct bio *bio) -{ - if (q->last_merge) - return elv_try_merge(q->last_merge, bio); - - return ELEVATOR_NO_MERGE; -} -EXPORT_SYMBOL(elv_try_last_merge); - static struct elevator_type *elevator_find(const char *name) { struct elevator_type *e = NULL; @@ -239,6 +230,9 @@ void elv_dispatch_sort(request_queue_t *q, struct request *rq) unsigned max_back; struct list_head *entry; + if (q->last_merge == rq) + q->last_merge = NULL; + boundary = q->end_sector; max_back = q->max_back_kb * 2; boundary = boundary > max_back ? boundary - max_back : 0; @@ -265,6 +259,15 @@ void elv_dispatch_sort(request_queue_t *q, struct request *rq) int elv_merge(request_queue_t *q, struct request **req, struct bio *bio) { elevator_t *e = q->elevator; + int ret; + + if (q->last_merge) { + ret = elv_try_merge(q->last_merge, bio); + if (ret != ELEVATOR_NO_MERGE) { + *req = q->last_merge; + return ret; + } + } if (e->ops->elevator_merge_fn) return e->ops->elevator_merge_fn(q, req, bio); @@ -278,6 +281,8 @@ void elv_merged_request(request_queue_t *q, struct request *rq) if (e->ops->elevator_merged_fn) e->ops->elevator_merged_fn(q, rq); + + q->last_merge = rq; } void elv_merge_requests(request_queue_t *q, struct request *rq, @@ -285,11 +290,10 @@ void elv_merge_requests(request_queue_t *q, struct request *rq, { elevator_t *e = q->elevator; - if (q->last_merge == next) - q->last_merge = NULL; - if (e->ops->elevator_merge_req_fn) e->ops->elevator_merge_req_fn(q, rq, next); + + q->last_merge = rq; } void elv_requeue_request(request_queue_t *q, struct request *rq) @@ -384,6 +388,8 @@ void __elv_add_request(request_queue_t *q, struct request *rq, int where, BUG_ON(!blk_fs_request(rq)); rq->flags |= REQ_SORTED; q->elevator->ops->elevator_add_req_fn(q, rq); + if (q->last_merge == NULL && rq_mergeable(rq)) + q->last_merge = rq; break; default: @@ -462,9 +468,6 @@ struct request *elv_next_request(request_queue_t *q) rq->flags |= REQ_STARTED; } - if (rq == q->last_merge) - q->last_merge = NULL; - if (!q->boundary_rq || q->boundary_rq == rq) { q->end_sector = rq_end_sector(rq); q->boundary_rq = NULL; @@ -518,16 +521,6 @@ void elv_dequeue_request(request_queue_t *q, struct request *rq) */ if (blk_account_rq(rq)) q->in_flight++; - - /* - * the main clearing point for q->last_merge is on retrieval of - * request by driver (it calls elv_next_request()), but it _can_ - * also happen here if a request is added to the queue but later - * deleted without ever being given to driver (merged with another - * request). - */ - if (rq == q->last_merge) - q->last_merge = NULL; } int elv_queue_empty(request_queue_t *q) diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index a2e496f285a4..bd55b1df8997 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -630,6 +630,9 @@ static inline void blkdev_dequeue_request(struct request *req) static inline void elv_dispatch_add_tail(struct request_queue *q, struct request *rq) { + if (q->last_merge == rq) + q->last_merge = NULL; + q->end_sector = rq_end_sector(rq); q->boundary_rq = rq; list_add_tail(&rq->queuelist, &q->queue_head); -- cgit v1.2.3 From cb19833dccb32f97cacbfff834b53523915f13f6 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Mon, 24 Oct 2005 08:35:58 +0200 Subject: [BLOCK] kill generic max_back_kb handling This patch kills max_back_kb handling from elv_dispatch_sort() and kills max_back_kb field from struct request_queue. Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- drivers/block/elevator.c | 6 +----- include/linux/blkdev.h | 1 - 2 files changed, 1 insertion(+), 6 deletions(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index 415144372c75..af2388e73f61 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -136,7 +136,6 @@ static int elevator_attach(request_queue_t *q, struct elevator_type *e, q->elevator = eq; q->end_sector = 0; q->boundary_rq = NULL; - q->max_back_kb = 0; if (eq->ops->elevator_init_fn) ret = eq->ops->elevator_init_fn(q, eq); @@ -227,16 +226,13 @@ void elevator_exit(elevator_t *e) void elv_dispatch_sort(request_queue_t *q, struct request *rq) { sector_t boundary; - unsigned max_back; struct list_head *entry; if (q->last_merge == rq) q->last_merge = NULL; boundary = q->end_sector; - max_back = q->max_back_kb * 2; - boundary = boundary > max_back ? boundary - max_back : 0; - + list_for_each_prev(entry, &q->queue_head) { struct request *pos = list_entry_rq(entry); diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index bd55b1df8997..159dbcd2eb59 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -339,7 +339,6 @@ struct request_queue */ sector_t end_sector; struct request *boundary_rq; - unsigned int max_back_kb; /* * Auto-unplugging state -- cgit v1.2.3 From cb98fc8bb9c141009e2bda99c0db39d387e142cf Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Fri, 28 Oct 2005 08:29:39 +0200 Subject: [BLOCK] Reimplement elevator switch This patch reimplements elevator switch. This patch assumes generic dispatch queue patchset is applied. * Each request is tagged with REQ_ELVPRIV flag if it has its elevator private data set. * Requests which doesn't have REQ_ELVPRIV flag set never enter iosched. They are always directly back inserted to dispatch queue. Of course, elevator_put_req_fn is called only for requests which have its REQ_ELVPRIV set. * Request queue maintains the current number of requests which have its elevator data set (elevator_set_req_fn called) in q->rq->elvpriv. * If a request queue has QUEUE_FLAG_BYPASS set, elevator private data is not allocated for new requests. To switch to another iosched, we set QUEUE_FLAG_BYPASS and wait until elvpriv goes to zero; then, we attach the new iosched and clears QUEUE_FLAG_BYPASS. New implementation is much simpler and main code paths are less cluttered, IMHO. Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- drivers/block/elevator.c | 78 ++++++++++++------------- drivers/block/ll_rw_blk.c | 142 +++++++++------------------------------------- include/linux/blkdev.h | 10 ++-- 3 files changed, 66 insertions(+), 164 deletions(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index af2388e73f61..272d93946621 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -34,6 +34,7 @@ #include #include #include +#include #include @@ -131,11 +132,7 @@ static int elevator_attach(request_queue_t *q, struct elevator_type *e, eq->ops = &e->ops; eq->elevator_type = e; - INIT_LIST_HEAD(&q->queue_head); - q->last_merge = NULL; q->elevator = eq; - q->end_sector = 0; - q->boundary_rq = NULL; if (eq->ops->elevator_init_fn) ret = eq->ops->elevator_init_fn(q, eq); @@ -184,6 +181,12 @@ int elevator_init(request_queue_t *q, char *name) struct elevator_queue *eq; int ret = 0; + INIT_LIST_HEAD(&q->queue_head); + q->last_merge = NULL; + q->end_sector = 0; + q->boundary_rq = NULL; + q->max_back_kb = 0; + elevator_setup_default(); if (!name) @@ -336,23 +339,14 @@ void __elv_add_request(request_queue_t *q, struct request *rq, int where, q->end_sector = rq_end_sector(rq); q->boundary_rq = rq; } - } + } else if (!(rq->flags & REQ_ELVPRIV) && where == ELEVATOR_INSERT_SORT) + where = ELEVATOR_INSERT_BACK; if (plug) blk_plug_device(q); rq->q = q; - if (unlikely(test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags))) { - /* - * if drain is set, store the request "locally". when the drain - * is finished, the requests will be handed ordered to the io - * scheduler - */ - list_add_tail(&rq->queuelist, &q->drain_list); - return; - } - switch (where) { case ELEVATOR_INSERT_FRONT: rq->flags |= REQ_SOFTBARRIER; @@ -659,25 +653,36 @@ EXPORT_SYMBOL_GPL(elv_unregister); * switch to new_e io scheduler. be careful not to introduce deadlocks - * we don't free the old io scheduler, before we have allocated what we * need for the new one. this way we have a chance of going back to the old - * one, if the new one fails init for some reason. we also do an intermediate - * switch to noop to ensure safety with stack-allocated requests, since they - * don't originate from the block layer allocator. noop is safe here, because - * it never needs to touch the elevator itself for completion events. DRAIN - * flags will make sure we don't touch it for additions either. + * one, if the new one fails init for some reason. */ static void elevator_switch(request_queue_t *q, struct elevator_type *new_e) { - elevator_t *e = kmalloc(sizeof(elevator_t), GFP_KERNEL); - struct elevator_type *noop_elevator = NULL; - elevator_t *old_elevator; + elevator_t *old_elevator, *e; + /* + * Allocate new elevator + */ + e = kmalloc(sizeof(elevator_t), GFP_KERNEL); if (!e) goto error; /* - * first step, drain requests from the block freelist + * Turn on BYPASS and drain all requests w/ elevator private data */ - blk_wait_queue_drained(q, 0); + spin_lock_irq(q->queue_lock); + + set_bit(QUEUE_FLAG_BYPASS, &q->queue_flags); + + while (q->elevator->ops->elevator_dispatch_fn(q, 1)) + ; + + while (q->rq.elvpriv) { + spin_unlock_irq(q->queue_lock); + msleep(100); + spin_lock_irq(q->queue_lock); + } + + spin_unlock_irq(q->queue_lock); /* * unregister old elevator data @@ -685,18 +690,6 @@ static void elevator_switch(request_queue_t *q, struct elevator_type *new_e) elv_unregister_queue(q); old_elevator = q->elevator; - /* - * next step, switch to noop since it uses no private rq structures - * and doesn't allocate any memory for anything. then wait for any - * non-fs requests in-flight - */ - noop_elevator = elevator_get("noop"); - spin_lock_irq(q->queue_lock); - elevator_attach(q, noop_elevator, e); - spin_unlock_irq(q->queue_lock); - - blk_wait_queue_drained(q, 1); - /* * attach and start new elevator */ @@ -707,11 +700,10 @@ static void elevator_switch(request_queue_t *q, struct elevator_type *new_e) goto fail_register; /* - * finally exit old elevator and start queue again + * finally exit old elevator and turn off BYPASS. */ elevator_exit(old_elevator); - blk_finish_queue_drain(q); - elevator_put(noop_elevator); + clear_bit(QUEUE_FLAG_BYPASS, &q->queue_flags); return; fail_register: @@ -720,13 +712,13 @@ fail_register: * one again (along with re-adding the sysfs dir) */ elevator_exit(e); + e = NULL; fail: q->elevator = old_elevator; elv_register_queue(q); - blk_finish_queue_drain(q); + clear_bit(QUEUE_FLAG_BYPASS, &q->queue_flags); + kfree(e); error: - if (noop_elevator) - elevator_put(noop_elevator); elevator_put(new_e); printk(KERN_ERR "elevator: switch to %s failed\n",new_e->elevator_name); } diff --git a/drivers/block/ll_rw_blk.c b/drivers/block/ll_rw_blk.c index d2a66fd309c3..f7c9931cb380 100644 --- a/drivers/block/ll_rw_blk.c +++ b/drivers/block/ll_rw_blk.c @@ -263,8 +263,6 @@ void blk_queue_make_request(request_queue_t * q, make_request_fn * mfn) blk_queue_bounce_limit(q, BLK_BOUNCE_HIGH); blk_queue_activity_fn(q, NULL, NULL); - - INIT_LIST_HEAD(&q->drain_list); } EXPORT_SYMBOL(blk_queue_make_request); @@ -1050,6 +1048,7 @@ static char *rq_flags[] = { "REQ_STARTED", "REQ_DONTPREP", "REQ_QUEUED", + "REQ_ELVPRIV", "REQ_PC", "REQ_BLOCK_PC", "REQ_SENSE", @@ -1640,9 +1639,9 @@ static int blk_init_free_list(request_queue_t *q) rl->count[READ] = rl->count[WRITE] = 0; rl->starved[READ] = rl->starved[WRITE] = 0; + rl->elvpriv = 0; init_waitqueue_head(&rl->wait[READ]); init_waitqueue_head(&rl->wait[WRITE]); - init_waitqueue_head(&rl->drain); rl->rq_pool = mempool_create_node(BLKDEV_MIN_RQ, mempool_alloc_slab, mempool_free_slab, request_cachep, q->node); @@ -1785,12 +1784,14 @@ EXPORT_SYMBOL(blk_get_queue); static inline void blk_free_request(request_queue_t *q, struct request *rq) { - elv_put_request(q, rq); + if (rq->flags & REQ_ELVPRIV) + elv_put_request(q, rq); mempool_free(rq, q->rq.rq_pool); } static inline struct request * -blk_alloc_request(request_queue_t *q, int rw, struct bio *bio, int gfp_mask) +blk_alloc_request(request_queue_t *q, int rw, struct bio *bio, + int priv, int gfp_mask) { struct request *rq = mempool_alloc(q->rq.rq_pool, gfp_mask); @@ -1803,11 +1804,15 @@ blk_alloc_request(request_queue_t *q, int rw, struct bio *bio, int gfp_mask) */ rq->flags = rw; - if (!elv_set_request(q, rq, bio, gfp_mask)) - return rq; + if (priv) { + if (unlikely(elv_set_request(q, rq, bio, gfp_mask))) { + mempool_free(rq, q->rq.rq_pool); + return NULL; + } + rq->flags |= REQ_ELVPRIV; + } - mempool_free(rq, q->rq.rq_pool); - return NULL; + return rq; } /* @@ -1863,22 +1868,18 @@ static void __freed_request(request_queue_t *q, int rw) * A request has just been released. Account for it, update the full and * congestion status, wake up any waiters. Called under q->queue_lock. */ -static void freed_request(request_queue_t *q, int rw) +static void freed_request(request_queue_t *q, int rw, int priv) { struct request_list *rl = &q->rq; rl->count[rw]--; + if (priv) + rl->elvpriv--; __freed_request(q, rw); if (unlikely(rl->starved[rw ^ 1])) __freed_request(q, rw ^ 1); - - if (!rl->count[READ] && !rl->count[WRITE]) { - smp_mb(); - if (unlikely(waitqueue_active(&rl->drain))) - wake_up(&rl->drain); - } } #define blkdev_free_rq(list) list_entry((list)->next, struct request, queuelist) @@ -1893,9 +1894,7 @@ static struct request *get_request(request_queue_t *q, int rw, struct bio *bio, struct request *rq = NULL; struct request_list *rl = &q->rq; struct io_context *ioc = current_io_context(GFP_ATOMIC); - - if (unlikely(test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags))) - goto out; + int priv; if (rl->count[rw]+1 >= q->nr_requests) { /* @@ -1940,9 +1939,14 @@ get_rq: rl->starved[rw] = 0; if (rl->count[rw] >= queue_congestion_on_threshold(q)) set_queue_congested(q, rw); + + priv = !test_bit(QUEUE_FLAG_BYPASS, &q->queue_flags); + if (priv) + rl->elvpriv++; + spin_unlock_irq(q->queue_lock); - rq = blk_alloc_request(q, rw, bio, gfp_mask); + rq = blk_alloc_request(q, rw, bio, priv, gfp_mask); if (!rq) { /* * Allocation failed presumably due to memory. Undo anything @@ -1952,7 +1956,7 @@ get_rq: * wait queue, but this is pretty rare. */ spin_lock_irq(q->queue_lock); - freed_request(q, rw); + freed_request(q, rw, priv); /* * in the very unlikely event that allocation failed and no @@ -2470,11 +2474,12 @@ static void __blk_put_request(request_queue_t *q, struct request *req) */ if (rl) { int rw = rq_data_dir(req); + int priv = req->flags & REQ_ELVPRIV; BUG_ON(!list_empty(&req->queuelist)); blk_free_request(q, req); - freed_request(q, rw); + freed_request(q, rw, priv); } } @@ -2802,97 +2807,6 @@ static inline void blk_partition_remap(struct bio *bio) } } -void blk_finish_queue_drain(request_queue_t *q) -{ - struct request_list *rl = &q->rq; - struct request *rq; - int requeued = 0; - - spin_lock_irq(q->queue_lock); - clear_bit(QUEUE_FLAG_DRAIN, &q->queue_flags); - - while (!list_empty(&q->drain_list)) { - rq = list_entry_rq(q->drain_list.next); - - list_del_init(&rq->queuelist); - elv_requeue_request(q, rq); - requeued++; - } - - if (requeued) - q->request_fn(q); - - spin_unlock_irq(q->queue_lock); - - wake_up(&rl->wait[0]); - wake_up(&rl->wait[1]); - wake_up(&rl->drain); -} - -static int wait_drain(request_queue_t *q, struct request_list *rl, int dispatch) -{ - int wait = rl->count[READ] + rl->count[WRITE]; - - if (dispatch) - wait += !list_empty(&q->queue_head); - - return wait; -} - -/* - * We rely on the fact that only requests allocated through blk_alloc_request() - * have io scheduler private data structures associated with them. Any other - * type of request (allocated on stack or through kmalloc()) should not go - * to the io scheduler core, but be attached to the queue head instead. - */ -void blk_wait_queue_drained(request_queue_t *q, int wait_dispatch) -{ - struct request_list *rl = &q->rq; - DEFINE_WAIT(wait); - - spin_lock_irq(q->queue_lock); - set_bit(QUEUE_FLAG_DRAIN, &q->queue_flags); - - while (wait_drain(q, rl, wait_dispatch)) { - prepare_to_wait(&rl->drain, &wait, TASK_UNINTERRUPTIBLE); - - if (wait_drain(q, rl, wait_dispatch)) { - __generic_unplug_device(q); - spin_unlock_irq(q->queue_lock); - io_schedule(); - spin_lock_irq(q->queue_lock); - } - - finish_wait(&rl->drain, &wait); - } - - spin_unlock_irq(q->queue_lock); -} - -/* - * block waiting for the io scheduler being started again. - */ -static inline void block_wait_queue_running(request_queue_t *q) -{ - DEFINE_WAIT(wait); - - while (unlikely(test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags))) { - struct request_list *rl = &q->rq; - - prepare_to_wait_exclusive(&rl->drain, &wait, - TASK_UNINTERRUPTIBLE); - - /* - * re-check the condition. avoids using prepare_to_wait() - * in the fast path (queue is running) - */ - if (test_bit(QUEUE_FLAG_DRAIN, &q->queue_flags)) - io_schedule(); - - finish_wait(&rl->drain, &wait); - } -} - static void handle_bad_sector(struct bio *bio) { char b[BDEVNAME_SIZE]; @@ -2988,8 +2902,6 @@ end_io: if (unlikely(test_bit(QUEUE_FLAG_DEAD, &q->queue_flags))) goto end_io; - block_wait_queue_running(q); - /* * If this device has partitions, remap block n * of partition p to block n+start(p) of the disk. diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 159dbcd2eb59..6186d5e2110f 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -107,9 +107,9 @@ typedef void (rq_end_io_fn)(struct request *); struct request_list { int count[2]; int starved[2]; + int elvpriv; mempool_t *rq_pool; wait_queue_head_t wait[2]; - wait_queue_head_t drain; }; #define BLK_MAX_CDB 16 @@ -211,6 +211,7 @@ enum rq_flag_bits { __REQ_STARTED, /* drive already may have started this one */ __REQ_DONTPREP, /* don't call prep for this one */ __REQ_QUEUED, /* uses queueing */ + __REQ_ELVPRIV, /* elevator private data attached */ /* * for ATA/ATAPI devices */ @@ -244,6 +245,7 @@ enum rq_flag_bits { #define REQ_STARTED (1 << __REQ_STARTED) #define REQ_DONTPREP (1 << __REQ_DONTPREP) #define REQ_QUEUED (1 << __REQ_QUEUED) +#define REQ_ELVPRIV (1 << __REQ_ELVPRIV) #define REQ_PC (1 << __REQ_PC) #define REQ_BLOCK_PC (1 << __REQ_BLOCK_PC) #define REQ_SENSE (1 << __REQ_SENSE) @@ -413,8 +415,6 @@ struct request_queue unsigned int sg_reserved_size; int node; - struct list_head drain_list; - /* * reserved for flush operations */ @@ -442,7 +442,7 @@ enum { #define QUEUE_FLAG_DEAD 5 /* queue being torn down */ #define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */ #define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */ -#define QUEUE_FLAG_DRAIN 8 /* draining queue for sched switch */ +#define QUEUE_FLAG_BYPASS 8 /* don't use elevator, just do FIFO */ #define QUEUE_FLAG_FLUSH 9 /* doing barrier flush sequence */ #define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags) @@ -668,8 +668,6 @@ extern void blk_dump_rq_flags(struct request *, char *); extern void generic_unplug_device(request_queue_t *); extern void __generic_unplug_device(request_queue_t *); extern long nr_blockdev_pages(void); -extern void blk_wait_queue_drained(request_queue_t *, int); -extern void blk_finish_queue_drain(request_queue_t *); int blk_get_queue(request_queue_t *); request_queue_t *blk_alloc_queue(int gfp_mask); -- cgit v1.2.3 From 64521d1a3ba7cc7fe4ab5640da83d88144f05340 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Fri, 28 Oct 2005 08:30:39 +0200 Subject: [BLOCK] elevator switch fixes/cleanup - 100msec sleep is a little excessive, lots of requests can complete in that timeframe. Use 10msec instead. - Rename QUEUE_FLAG_BYPASS to QUEUE_FLAG_ELVSWITCH to indicate what is going on. Signed-off-by: Jens Axboe --- drivers/block/elevator.c | 8 ++++---- drivers/block/ll_rw_blk.c | 2 +- include/linux/blkdev.h | 2 +- 3 files changed, 6 insertions(+), 6 deletions(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index 272d93946621..cf9bedbb761b 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -671,14 +671,14 @@ static void elevator_switch(request_queue_t *q, struct elevator_type *new_e) */ spin_lock_irq(q->queue_lock); - set_bit(QUEUE_FLAG_BYPASS, &q->queue_flags); + set_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags); while (q->elevator->ops->elevator_dispatch_fn(q, 1)) ; while (q->rq.elvpriv) { spin_unlock_irq(q->queue_lock); - msleep(100); + msleep(10); spin_lock_irq(q->queue_lock); } @@ -703,7 +703,7 @@ static void elevator_switch(request_queue_t *q, struct elevator_type *new_e) * finally exit old elevator and turn off BYPASS. */ elevator_exit(old_elevator); - clear_bit(QUEUE_FLAG_BYPASS, &q->queue_flags); + clear_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags); return; fail_register: @@ -716,7 +716,7 @@ fail_register: fail: q->elevator = old_elevator; elv_register_queue(q); - clear_bit(QUEUE_FLAG_BYPASS, &q->queue_flags); + clear_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags); kfree(e); error: elevator_put(new_e); diff --git a/drivers/block/ll_rw_blk.c b/drivers/block/ll_rw_blk.c index f7c9931cb380..fc12d019b19d 100644 --- a/drivers/block/ll_rw_blk.c +++ b/drivers/block/ll_rw_blk.c @@ -1940,7 +1940,7 @@ get_rq: if (rl->count[rw] >= queue_congestion_on_threshold(q)) set_queue_congested(q, rw); - priv = !test_bit(QUEUE_FLAG_BYPASS, &q->queue_flags); + priv = !test_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags); if (priv) rl->elvpriv++; diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 6186d5e2110f..7d9d14d6d8f9 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -442,7 +442,7 @@ enum { #define QUEUE_FLAG_DEAD 5 /* queue being torn down */ #define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */ #define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */ -#define QUEUE_FLAG_BYPASS 8 /* don't use elevator, just do FIFO */ +#define QUEUE_FLAG_ELVSWITCH 8 /* don't use elevator, just do FIFO */ #define QUEUE_FLAG_FLUSH 9 /* doing barrier flush sequence */ #define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags) -- cgit v1.2.3 From 772eca7825358f52cf208afd7ab8ce22722ce41a Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Fri, 28 Oct 2005 09:01:17 +0200 Subject: [BLOCK] Leftover reference to ->max_back_kb Signed-off-by: Jens Axboe --- drivers/block/elevator.c | 1 - 1 file changed, 1 deletion(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index cf9bedbb761b..8de77416bd2f 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -185,7 +185,6 @@ int elevator_init(request_queue_t *q, char *name) q->last_merge = NULL; q->end_sector = 0; q->boundary_rq = NULL; - q->max_back_kb = 0; elevator_setup_default(); -- cgit v1.2.3 From 8267e268e0914ac9371d07f711fcf20cc572993c Mon Sep 17 00:00:00 2001 From: Al Viro Date: Fri, 21 Oct 2005 03:20:53 -0400 Subject: [PATCH] gfp_t: block layer core Signed-off-by: Al Viro Signed-off-by: Linus Torvalds --- drivers/block/as-iosched.c | 2 +- drivers/block/cfq-iosched.c | 8 ++++---- drivers/block/deadline-iosched.c | 2 +- drivers/block/elevator.c | 2 +- drivers/block/ll_rw_blk.c | 16 ++++++++-------- include/linux/blkdev.h | 14 +++++++------- include/linux/elevator.h | 4 ++-- 7 files changed, 24 insertions(+), 24 deletions(-) (limited to 'drivers/block/elevator.c') diff --git a/drivers/block/as-iosched.c b/drivers/block/as-iosched.c index 95c0a3690b0f..1f08e14697e9 100644 --- a/drivers/block/as-iosched.c +++ b/drivers/block/as-iosched.c @@ -1807,7 +1807,7 @@ static void as_put_request(request_queue_t *q, struct request *rq) } static int as_set_request(request_queue_t *q, struct request *rq, - struct bio *bio, int gfp_mask) + struct bio *bio, gfp_t gfp_mask) { struct as_data *ad = q->elevator->elevator_data; struct as_rq *arq = mempool_alloc(ad->arq_pool, gfp_mask); diff --git a/drivers/block/cfq-iosched.c b/drivers/block/cfq-iosched.c index cd056e7e64ec..d3bfe8cfb039 100644 --- a/drivers/block/cfq-iosched.c +++ b/drivers/block/cfq-iosched.c @@ -1422,7 +1422,7 @@ static void cfq_exit_io_context(struct cfq_io_context *cic) } static struct cfq_io_context * -cfq_alloc_io_context(struct cfq_data *cfqd, int gfp_mask) +cfq_alloc_io_context(struct cfq_data *cfqd, gfp_t gfp_mask) { struct cfq_io_context *cic = kmem_cache_alloc(cfq_ioc_pool, gfp_mask); @@ -1517,7 +1517,7 @@ static int cfq_ioc_set_ioprio(struct io_context *ioc, unsigned int ioprio) static struct cfq_queue * cfq_get_queue(struct cfq_data *cfqd, unsigned int key, unsigned short ioprio, - int gfp_mask) + gfp_t gfp_mask) { const int hashval = hash_long(key, CFQ_QHASH_SHIFT); struct cfq_queue *cfqq, *new_cfqq = NULL; @@ -1578,7 +1578,7 @@ out: * cfqq, so we don't need to worry about it disappearing */ static struct cfq_io_context * -cfq_get_io_context(struct cfq_data *cfqd, pid_t pid, int gfp_mask) +cfq_get_io_context(struct cfq_data *cfqd, pid_t pid, gfp_t gfp_mask) { struct io_context *ioc = NULL; struct cfq_io_context *cic; @@ -2075,7 +2075,7 @@ static void cfq_put_request(request_queue_t *q, struct request *rq) */ static int cfq_set_request(request_queue_t *q, struct request *rq, struct bio *bio, - int gfp_mask) + gfp_t gfp_mask) { struct cfq_data *cfqd = q->elevator->elevator_data; struct task_struct *tsk = current; diff --git a/drivers/block/deadline-iosched.c b/drivers/block/deadline-iosched.c index 52a3ae5289a0..753546ba2262 100644 --- a/drivers/block/deadline-iosched.c +++ b/drivers/block/deadline-iosched.c @@ -756,7 +756,7 @@ static void deadline_put_request(request_queue_t *q, struct request *rq) static int deadline_set_request(request_queue_t *q, struct request *rq, struct bio *bio, - int gfp_mask) + gfp_t gfp_mask) { struct deadline_data *dd = q->elevator->elevator_data; struct deadline_rq *drq; diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index 98f0126a2deb..c744d2a13062 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -487,7 +487,7 @@ struct request *elv_former_request(request_queue_t *q, struct request *rq) } int elv_set_request(request_queue_t *q, struct request *rq, struct bio *bio, - int gfp_mask) + gfp_t gfp_mask) { elevator_t *e = q->elevator; diff --git a/drivers/block/ll_rw_blk.c b/drivers/block/ll_rw_blk.c index baedac522945..0f64ee7d8d27 100644 --- a/drivers/block/ll_rw_blk.c +++ b/drivers/block/ll_rw_blk.c @@ -1652,13 +1652,13 @@ static int blk_init_free_list(request_queue_t *q) static int __make_request(request_queue_t *, struct bio *); -request_queue_t *blk_alloc_queue(int gfp_mask) +request_queue_t *blk_alloc_queue(gfp_t gfp_mask) { return blk_alloc_queue_node(gfp_mask, -1); } EXPORT_SYMBOL(blk_alloc_queue); -request_queue_t *blk_alloc_queue_node(int gfp_mask, int node_id) +request_queue_t *blk_alloc_queue_node(gfp_t gfp_mask, int node_id) { request_queue_t *q; @@ -1787,7 +1787,7 @@ static inline void blk_free_request(request_queue_t *q, struct request *rq) } static inline struct request * -blk_alloc_request(request_queue_t *q, int rw, struct bio *bio, int gfp_mask) +blk_alloc_request(request_queue_t *q, int rw, struct bio *bio, gfp_t gfp_mask) { struct request *rq = mempool_alloc(q->rq.rq_pool, gfp_mask); @@ -1885,7 +1885,7 @@ static void freed_request(request_queue_t *q, int rw) * Returns !NULL on success, with queue_lock *not held*. */ static struct request *get_request(request_queue_t *q, int rw, struct bio *bio, - int gfp_mask) + gfp_t gfp_mask) { struct request *rq = NULL; struct request_list *rl = &q->rq; @@ -2019,7 +2019,7 @@ static struct request *get_request_wait(request_queue_t *q, int rw, return rq; } -struct request *blk_get_request(request_queue_t *q, int rw, int gfp_mask) +struct request *blk_get_request(request_queue_t *q, int rw, gfp_t gfp_mask) { struct request *rq; @@ -2251,7 +2251,7 @@ EXPORT_SYMBOL(blk_rq_unmap_user); * @gfp_mask: memory allocation flags */ int blk_rq_map_kern(request_queue_t *q, struct request *rq, void *kbuf, - unsigned int len, unsigned int gfp_mask) + unsigned int len, gfp_t gfp_mask) { struct bio *bio; @@ -3393,7 +3393,7 @@ void exit_io_context(void) * but since the current task itself holds a reference, the context can be * used in general code, so long as it stays within `current` context. */ -struct io_context *current_io_context(int gfp_flags) +struct io_context *current_io_context(gfp_t gfp_flags) { struct task_struct *tsk = current; struct io_context *ret; @@ -3424,7 +3424,7 @@ EXPORT_SYMBOL(current_io_context); * * This is always called in the context of the task which submitted the I/O. */ -struct io_context *get_io_context(int gfp_flags) +struct io_context *get_io_context(gfp_t gfp_flags) { struct io_context *ret; ret = current_io_context(gfp_flags); diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index efdc9b5bc05c..1afbdb2d752c 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -96,8 +96,8 @@ struct io_context { void put_io_context(struct io_context *ioc); void exit_io_context(void); -struct io_context *current_io_context(int gfp_flags); -struct io_context *get_io_context(int gfp_flags); +struct io_context *current_io_context(gfp_t gfp_flags); +struct io_context *get_io_context(gfp_t gfp_flags); void copy_io_context(struct io_context **pdst, struct io_context **psrc); void swap_io_context(struct io_context **ioc1, struct io_context **ioc2); @@ -354,7 +354,7 @@ struct request_queue * queue needs bounce pages for pages above this limit */ unsigned long bounce_pfn; - unsigned int bounce_gfp; + gfp_t bounce_gfp; /* * various queue flags, see QUEUE_* below @@ -550,7 +550,7 @@ extern void generic_make_request(struct bio *bio); extern void blk_put_request(struct request *); extern void blk_end_sync_rq(struct request *rq); extern void blk_attempt_remerge(request_queue_t *, struct request *); -extern struct request *blk_get_request(request_queue_t *, int, int); +extern struct request *blk_get_request(request_queue_t *, int, gfp_t); extern void blk_insert_request(request_queue_t *, struct request *, int, void *); extern void blk_requeue_request(request_queue_t *, struct request *); extern void blk_plug_device(request_queue_t *); @@ -565,7 +565,7 @@ extern void blk_run_queue(request_queue_t *); extern void blk_queue_activity_fn(request_queue_t *, activity_fn *, void *); extern int blk_rq_map_user(request_queue_t *, struct request *, void __user *, unsigned int); extern int blk_rq_unmap_user(struct bio *, unsigned int); -extern int blk_rq_map_kern(request_queue_t *, struct request *, void *, unsigned int, unsigned int); +extern int blk_rq_map_kern(request_queue_t *, struct request *, void *, unsigned int, gfp_t); extern int blk_rq_map_user_iov(request_queue_t *, struct request *, struct sg_iovec *, int); extern int blk_execute_rq(request_queue_t *, struct gendisk *, struct request *, int); @@ -654,8 +654,8 @@ extern void blk_wait_queue_drained(request_queue_t *, int); extern void blk_finish_queue_drain(request_queue_t *); int blk_get_queue(request_queue_t *); -request_queue_t *blk_alloc_queue(int gfp_mask); -request_queue_t *blk_alloc_queue_node(int,int); +request_queue_t *blk_alloc_queue(gfp_t); +request_queue_t *blk_alloc_queue_node(gfp_t, int); #define blk_put_queue(q) blk_cleanup_queue((q)) /* diff --git a/include/linux/elevator.h b/include/linux/elevator.h index ea6bbc2d7407..ed93125c1db5 100644 --- a/include/linux/elevator.h +++ b/include/linux/elevator.h @@ -18,7 +18,7 @@ typedef struct request *(elevator_request_list_fn) (request_queue_t *, struct re typedef void (elevator_completed_req_fn) (request_queue_t *, struct request *); typedef int (elevator_may_queue_fn) (request_queue_t *, int, struct bio *); -typedef int (elevator_set_req_fn) (request_queue_t *, struct request *, struct bio *, int); +typedef int (elevator_set_req_fn) (request_queue_t *, struct request *, struct bio *, gfp_t); typedef void (elevator_put_req_fn) (request_queue_t *, struct request *); typedef void (elevator_deactivate_req_fn) (request_queue_t *, struct request *); @@ -98,7 +98,7 @@ extern int elv_register_queue(request_queue_t *q); extern void elv_unregister_queue(request_queue_t *q); extern int elv_may_queue(request_queue_t *, int, struct bio *); extern void elv_completed_request(request_queue_t *, struct request *); -extern int elv_set_request(request_queue_t *, struct request *, struct bio *, int); +extern int elv_set_request(request_queue_t *, struct request *, struct bio *, gfp_t); extern void elv_put_request(request_queue_t *, struct request *); /* -- cgit v1.2.3